- Email: [email protected]
Airborne inhalable metals in residential areas of Delhi, India: distribution, source apportionment and health risks
Atmospheric Pollution Research 3 (2012) 46‐54
Atmospheric Pollution Research
www.atmospolres.com
Airborne inhalable metals in residential areas of Delhi, India: distribution, source apportionment and health risks Pandit S. Khillare, Sayantan Sarkar Environmental Monitoring and Assessment Laboratory, School of Environmental Sciences, Jawaharlal Nehru University, New Delhi – 110067, India
ABSTRACT
The inhalable fraction of ambient particles (particulate matter with aerodynamic diameter less than 10 µm, denoted by PM10) collected at three residential sites in Delhi, India during December 2008–November 2009 were characterized with respect to 8 major and trace metals (Fe, Mn, Cd, Cu, Ni, Pb, Zn and Cr). Mean annual 24–h PM10 levels varied from –3 166.5–192.3 µg m at the sites (8–10 times of the WHO limit). Weekday/weekend effects on PM10 and associated metals were investigated. Significant seasonal variations (ANOVA, p<0.001) in species concentrations were observed with PM10 and crustal metals peaking in summer while anthropogenic metals peaked in winter. Spatial distributions of metals were influenced mainly by proximity to traffic and industrial areas. Influence of meteorological parameters on ambient metal distributions was also studied. Source apportionment by principal component analysis–multiple linear regression (PCA–MLR) identified three major sources: crustal (49–65%), vehicular (27–31%) and industrial (4–21%). As a part of health risk assessment, calculated theoretical estimates of blood lead levels in children varied from 10.2– –1 13.3 µg dL , indicating some degree of lead toxicity. Incremental lifetime cancer risk (ILCR) assessment showed that up to 2 908 excess cancer cases (102 for Cd, 2 559 for Cr (VI) and 247 for Ni) are likely in Delhi for lifetime inhalation exposure to these species at their current concentrations.
Keywords: PM10 Metals Source apportionment PCA–MLR Health risk Delhi
Article History: Received: 13 June 2011 Revised: 31 August 2011 Accepted: 02 September 2011
Corresponding Author: Pandit S. Khillare Tel: +91‐11‐26704325 E‐mail: [email protected] © Author(s) 2012. This work is distributed under the Creative Commons Attribution 3.0 License.
doi: 10.5094/APR.2012.004
1. Introduction It has been estimated that urban air pollution causes 360 000 premature deaths each year in Asia (Stone et al., 2010) while an estimated 10 500 premature deaths occur per year in Delhi, the capital of India, for the same reason (Gurjar et al., 2010). Over the past decade, a large number of epidemiological studies have established a definite link between airborne particles and human mortality and/or morbidity (Schwartz et al., 1996; Pope and Dockery, 2006). Recent studies have put forth the opinion that chemical composition of particulates, especially the presence of small amounts of toxic chemicals, is a more appropriate parameter for the assessment of particulate health effects rather than their mass concentrations (Forsberg et al., 2005). In this context, metallic species are a class of atmospheric pollutants that are key determinants of the health effects of particulate matter. Some of these metals are known to induce formation of reactive oxygen species (ROS) and hydroxyl radicals (•OH) in biological tissues (de Kok et al., 2006; Nawrot et al., 2009) while others can trigger neurological disorders, different forms of cancer and other ailments (Canfield et al., 2003; IARC, 2009). Effective and cost– efficient regulation of these toxic metals in the atmosphere is not always achieved with a general reduction in aerosol concentrations (von Schneidemesser et al., 2010); proper identification and quan– tification of their sources and a comprehensive assessment of the risks they pose to human health are essential in prioritizing miti– gation requirements. With these objectives in mind we aimed at characterizing ambient inhalable particles (PM10, aerodynamic diameter ≤10 µm) with respect to 8 major and trace metals (Fe, Mn, Cd, Cu, Ni, Pb, Zn
and Cr) at three residential areas of Delhi. Previous studies from this area have generally focused on “pollution hot spots” such as traffic, industrial and commercial sites and, consequently, under– standing of the pollution scenario in typical residential areas of Delhi is fairly weak. Pollution assessment in residential areas is important since outdoor air quality has a major influence on indoor pollution. Population residence time is also highest in residential areas which results in greater exposure. The major sources of atmospheric pollution in Delhi include 129 000 industrial units located in 31 authorized industrial estates and several non– conforming areas (mostly within the city), three coal–fired power plants (combined capacity 1 100 MW) and a staggering vehicular population of 6.6 million. As such, land–use patterns in Delhi are overlapping and essentially mixed, which means that residential areas are likely to be impacted by a host of different sources that needs to be delineated. Overall, this study aims to present a holistic view of the distribution, spatio–temporal variation and source apportionment of metallic species at some residential areas of Delhi along with an estimation of possible health risks to the exposed population.
2. Materials and Methods 2.1. Study area Delhi (28°37’N, 77°12’E, population 17 million) is spread over an area of 1 483 km2 with an average elevation of approximately 216 m above sea level. The region is characterized by intensely hot summers (monthly mean temperatures of 32–34 °C in May–June) and cold winters (monthly mean temperatures of 12–14 °C in December–January). The mean annual rainfall is 714 mm, of which
Khillare and Sarkar – Atmospheric Pollution Research 3 (2012) 46‐54
around 80% is received during the monsoon months (July to October). The predominant wind direction is from the north and northwest (Figure 1) except during the monsoon season that is characterized by easterly or southeasterly winds.
2.2. Description of sampling sites Locations of the sampling sites are shown in Figure 1. Site 1, Rajghat (RG), is located in central Delhi lying 1–2 km northwest of Rajghat and Indraprastha coal–fired power plants and 8–10 km east–southeast of a cluster of industrial areas (Wazirpur, Jahangirpuri, Mangolpuri, Naraina etc.). A traffic intersection is present at a distance of 800 m. The sampler was located on the roof of a government building at about 8 m from the ground. Site 2, Mayur Vihar (MV), is located 4–4.5 km southeast of Rajghat and Indraprastha power plants and 600 m away from the Noida Link Road. The sampler was located on the roof of Ahlcon International School at a height of around 12–13 m. The Patparganj Industrial Area is located around 3 km to the northeast. Site 3, Mithapur (MP), is a suburban residential area located 2.5 km southeast of Badarpur power plant and 5–7 km southeast of Okhla Industrial Area (Phase I, II and III). The heavy–traffic Delhi–Mathura highway is at a distance of about 1.5 km. The sampler was located on the roof of a private household building at a height of approximately 12 m. 2.3. Sampling protocol PM10 samples were collected simultaneously at the three sites for a period of one year (December 2008–November 2009). The sampling frequency was once a week and sampling duration for each sample was 24 hours (10 AM to 10 AM the following day). Seasonal distribution of samples was as follows: winter –15 each at RG, MV and MP; summer –13 at RG, 16 at MV and 15 at MP; monsoon –15 at RG, 13 at MV and 14 at MP. A number of samples were lost due to logistical problems such as electrical failure, sampler malfunction etc. A total of 131 samples (43 at RG and 44 each at MV and MP) were collected over the course of this study. Around 63% of the sampling days were weekdays (Monday– Friday) while the remaining 37% were weekends (Saturday– Sunday). PM10 was trapped on Whatman GF/A (8” × 10”) glass fiber
47
filters (precombusted at 450°C for 12 hours) using High–volume samplers (Respirable Dust Sampler, Model MBLRDS–002, Mars 3 –1 Bioanalytical Pvt. Ltd.) having a constant flow rate of 1.2 m min . High–volume sampling with glass fiber filters is widely used for atmospheric metal determination (Vijayanand et al., 2008; Shridhar et al., 2010; Paulino et al., 2010; Yatkin and Bayram, 2010). Filters were transported to and from the field in sealed polyethylene bags and were desiccated for 48 hrs before and after use. PM10 loads were determined gravimetrically by weighing the filters twice in a microbalance (Model AE163, Mettler, sensitivity 0.0001 g) after proper conditioning. Filters were stored in a refrigerator (4 °C) until analysis. 2.4. Extraction and analysis of metals Metal extraction was carried out using a microwave digestion system (Speedwave MWS–3+, Berghof, Germany) consisting of a 2 450 MHz microwave power system with a maximum power of 1 450 W. It houses 12 DAP–100 TFMTM pressure vessels equipped with temperature and pressure sensors, screw caps and rupture discs. The maximum operable temperature and pressure are 230°C and 40 bars, respectively. Two circular portions (4 cm–diameter each) cut from diagonal locations on each filter paper using a stainless steel punch were digested with a mixed acid solution (10 mL HNO3 + 3 mL HF) according to the following program: (i) temperature ramped to 150°C within 10 min with a dwell time of 5 min, (ii) a further ramp to 190°C within 5 min with a dwell time of 5 min, (iii) temperature reduced to 100°C within 5 min with a dwell time of 15 min. The pressure was kept at 25 bars. The vessels were allowed to cool for 30 min and were vented and opened in a fume hood. After subse– quent filtration and volume make–up to 25 ml with high–purity deionized water (18.2 MΩ cm–1, Milli–Q system, Millipore, USA), the extracted solutions were transferred to polypropylene bottles and refrigerated until analysis. Metal analysis was carried out using a flame atomic absorption spectrometry system (Solaar M Series, Thermo Scientific) under optimized conditions. Metal recoveries were checked by spiking samples with standard solutions prior to extraction.
Figure 1. Map showing the sampling sites and wind rose showing the prevalent wind direction in Delhi during the study period (December 2008 – November 2009).
Khillare and Sarkar – Atmospheric Pollution Research 3 (2012) 46‐54
48
2.5. Analytical quality control The AA spectrometer was calibrated for each metal individ– ually using sets of five standards (Reagecon, Ireland) covering the range of concentrations encountered in ambient air work. The calibration curves were linear in the concentration range with linear regression coefficients (R2)>0.995 for linear least–squares fit of data. Samples were analyzed in triplicates to ensure precision. Values reported are means of such triplicate analyses. Relative standard deviations (RSDs) of replicate samples were less than 10%. To check homogeneity we separately analyzed three samples (each comprising of two 4 cm–diameter circles located diagonally of each other) taken from different spots on the same filter and found that RSDs for the analyzed metals were <17% in such replicates. Field blanks (around 20% of the total number of samples) and reagent blanks (one for each batch of samples) were analyzed to determine analytical bias. The data were subsequently blank–corrected as required. Detection limits (3σ) of metals were 5.2, 1.4, 0.6, 1.7, 5.4, 8, 0.8 and 3.1 ng m–3 for Fe, Mn, Cd, Cu, Ni, Pb, Zn and Cr, respectively. Metal recoveries ranged from 76–97%.
3. Results and Discussion 3.1. PM10 mass concentrations Mean annual PM10 at RG, MV and MP were 166.5 ± 54.7, 175.5 ± 67.6 and 192.3 ± 63.4 µg m–3, respectively (Table 1), giving –3 an overall mean of 178.2 ± 62.7 µg m for Delhi. These values are 3 times the annual PM10 National Ambient Air Quality Standard (NAAQS) (60 µg m–3) prescribed by the Central Pollution Control Board (CPCB) of India (MoEF, 2009) and around 8–10 times the annual PM10 air quality guideline (AQG) (20 µg m–3) set by the World Health Organization (WHO, 2006). The 24–h PM10 NAAQS –3 (100 µg m ) was violated on 88%, 89% and 93% of the sampling days at RG, MV and MP, respectively. The overall mean PM10 con– centration (mean of three sites) on weekdays (186.4 ± 41.8 µg m–3) –3 was greater than that on weekends (160.9 ± 70.2 µg m ) but the difference was not statistically significant (independent samples t– test, t = 1.55, p = 0.13). Assuming industrial and domestic emissions do not show day–of–the–week variations, higher PM10 concentrations on weekdays could be attributed to increased traffic. Spatial variations of PM10 were not statistically significant (F = 2.2, p = 0.12, Table 1), which probably suggests that the study sites were impacted by similar sources. 3.2. Loadings and spatial variation of ambient metals Descriptive statistics of PM10–associated metal concentrations are presented in Table 1. The most abundant metals were Fe and –3 Zn with concentrations in the range of several µg m whereas Cd was the least abundant with concentration ranges of several ng m–3. A comparison has been made between the metal concentrations observed in this study and those reported from
other parts of India and the world (Singh et al., 2002; Heal et al., 2005; Sanchez de la Campa et al., 2007; Meresova et al., 2008; Vijayanand et al., 2008; Kulshrestha et al., 2009; Jorquera, 2009; Limbeck et al., 2009; Contini et al., 2010; Khan et al., 2010; Shridhar et al., 2010; Theodosi et al., 2010; von Schneidemesser et al., 2010) in Table 2. It is evident that concentrations of inhalable metals in residential areas of Delhi – although comparable, in some cases, with other Indian and Asian cities – are often more than an order of magnitude greater than their European or US counter– parts. Metal concentrations obtained in this study were also compared with those observed at an urban background site in Delhi (Jawaharlal Nehru University campus, JNU) for the same time period (Sarkar and Khillare, submitted for publication). JNU is an institutional area in south Delhi situated on a ridge of the Aravalli Hills in the midst of a 4 km2 forested area. It is widely regarded as a clean and sensitive area of Delhi (Srivastava and Jain, 2008). Nearest traffic and industrial areas are at least 2 km and 8–9 km away, respectively. Concentrations of PM10 and a majority of the metals at JNU were found to be lower than RG, MV and MP by the following amounts: PM10 (28–37%), Fe (3–30%), Cd (50–79%), Pb (41–65%), Zn (24–34%) and Cr (46–70%). Manganese, copper and nickel concentrations at JNU were more or less comparable with those at RG, MV and MP. Overall, it was observed that concen– trations of PM10 and a number of metals at the chosen residential areas of Delhi were distinctly higher than those at the urban background site. Inter–site variations were significant for all metals except Zn (F = 1.8, p = 0.2). Absence of significant spatial variations in Zn concentration probably indicates ubiquitous traffic sources near the sampling sites. In fact, Cu/Zn ratios at RG, MV and MP were 0.04 ± 0.04, 0.07 ± 0.04 and 0.06 ± 0.05, respectively, all of which strongly suggest vehicular sources (Cadle et al., 1999). Spatial variations were in general more pronounced for anthropogenic metals than their crustal counterparts. Concentrations of Cd, Ni and Pb at MP were significantly higher (p<0.05) than RG. Proximity to a major coal–fired power plant and an industrial area could be the reasons for higher metal concentrations at MP. Being a suburban residential area, domestic emissions (in the form of charcoal, biomass and low–grade coal combustion) could also be significant. In comparison, RG is a relatively cleaner site lying upwind of power plants and having fewer emission sources in the vicinity. Comparatively lower concentrations of anthropogenic metals at this site are hence expected. The other site, MV, exhibited metal concentrations that were in many cases intermediate between and not significantly different from those at RG and MP. Overall, we noted that the differences in metal concentrations between the three investigated sites, although statistically significant in most cases, were not extremely high. This is expected to some extent because all three sites were typically residential in nature with similar land–use patterns and were possibly impacted by similar sources. Concentrations of traffic tracers (Cu, Zn and Cr) were on an average 3–14% higher on –3
Table 1. Seasonal and spatial distribution of ambient PM10 and associated metals (µg m ). Also shown are the results of two–way ANOVA test (F–values along with corresponding significance levels) Species PM10 Fe Mn Cd Cu Ni Pb Zn Cr
RG
W 165.1 7 0.34 0.014 0.28 0.44 0.41 4.4 0.17
S 205.8 12.1 0.37 0.005 0.16 0.32 0.24 4.4 0.17
M 133.8 5.5 0.26 0.005 0.08 0.15 0.14 5.3 0.07
W 184.7 9 0.27 0.022 0.25 0.39 0.47 5.3 0.32
MV S 210.3 12.4 0.37 0.014 0.34 0.3 0.49 4.7 0.17
M 122 6.7 0.21 0.011 0.22 0.13 0.25 2.7 0.09
W 194.9 10.5 0.22 0.028 0.26 0.46 0.51 5.7 0.32
MP S 215.9 13 0.32 0.018 0.21 0.39 0.54 4 0.27
M 164.2 10 0.2 0.011 0.16 0.25 0.33 2.6 0.09
RG (N = 43)
MV (N = 44) MP (N = 44)
Mean ± SD 166.5 ± 54.7 8 ± 3.7 0.32 ± 0.1 0.008 ± 0.006 0.18 ± 0.1 0.3 ± 0.2 0.27 ± 0.2 4.7 ± 1.7 0.13 ± 0.07
Mean ± SD Mean ± SD 175.5 ± 67.6 192.3 ± 63.4 9.6 ± 4.4 11.2 ± 4.3 0.29 ± 0.1 0.25 ± 0.1 0.016 ± 0.01 0.019 ± 0.015 0.27 ± 0.1 0.21 ± 0.1 0.28 ± 0.2 0.37 ± 0.2 0.41 ± 0.2 0.46 ± 0.3 4.3 ± 1.5 4.1 ± 2.1 0.2 ± 0.12 0.23 ± 0.18
F–values two–way ANOVA Season 17.2c 23.2c 14.2c 19.2c 12.3c 49c 17.8c 11.7c 36.3c
Site 2.2NS 7.7b 4.5a 16.1c 8.6c 7c 13.3c 1.8NS 8.7c
Season*Site 0.6NS 1.2NS 0.7NS 0.7NS 4.5b 0.5NS 1.5NS 8c 3.2b
W, S and M denote winter (November – February), summer (March – June) and monsoon (July – October) seasons, respectively; Mean ± SD refers to the annual mean values ± one standard deviation at the sites; N denotes the number of samples collected at a particular site; Level of significance: NS a b c not significant, p<0.05, p<0.01, p<0.001.
Khillare and Sarkar – Atmospheric Pollution Research 3 (2012) 46‐54
49
–3
Table 2. Comparison of ambient metal concentrations (µg m ) observed in the present study with those from other parts of India and the world Location/Site type/ Fe PM type Delhi, India/R/PM10 8–11.2 Coimbatore, India/R/SPM 2.9–6 Agra, India/U/PM10 2.9 16.4 Delhi, India/U/SPM 8.2 Lahore, Pakistan/U/PM10 Beijing, China/R/PM10 4.6 3.5 Tocopilla, Chile/R/PM10 Istanbul, Turkey/U/PM10 0.7 0.31 Lecce, Italy/UB/PM10 0.78 Vienna, Austria/U/PM10 Bratislava, Slovakia/U/NR 0.25 0.7 Huelva, Spain/U/PM10 Edinburgh, UK/UB/PM10 0.18 Los Angeles, USA/U/PM10 0.15
Mn
Cd
Cu
Ni
Pb
Zn
0.25–0.32 NR 0.9 0.74 0.3 0.21 0.06 0.02 0.008 0.01 0.005 0.01 0.003 0.002
0.01–0.02 BDL NR 0.01 0.08 0.005 NR 0.001 NR 0.0005 0.0001 0.0006 0.0003 NR
0.18–0.27 0.7–0.77 0.1 3.7 0.07 0.09 1.1 0.02 0.01 0.02 0.008 0.05 0.005 0.003
0.28–0.37 0.16–0.22 0.2 0.15 0.02 0.01 0.0009 0.007 0.003 0.006 0.0005 0.004 0.003 0.0006
0.27–0.46 0.21–0.62 1.1 0.44 4.4 0.33 0.01 0.07 0.008 0.01 0.02 0.02 0.01 0.002
4.1–4.7 11.3–20.7 0.5 4.7 11 0.64 0.01 0.24 0.03 0.04 0.03 0.03 0.01 0.004
Cr
Reference
0.13–0.23 Present study 0.005–0.88 Vijayanand et al., (2008) 0.3 Kulshrestha et al., (2009) 0.35 Shridhar et al., (2010) 0.03 von Schneidemesser et al., (2010) 0.02 Khan et al., (2010) NR Jorquera, (2009)a 0.004 Theodosi et al., (2010) 0.002 Contini et al., (2010) 0.006 Limbeck et al., (2009) 0.001 Meresova et al., (2008) 0.003 Sanchez de la Campa et al., (2007) 0.002 Heal et al., (2005)b 0.001 Singh et al., (2002)
R: residential site, U: urban site, UB: urban background site; NR: not reported; BDL: below detection limit a sampling in March–April 2006 b median values
weekdays as compared to weekends, but the differences were not statistically significant (p>0.05). This possibly suggests that traffic volumes were not significantly different at the sites between weekdays and weekends. 3.3. Enrichment Factors of metals Enrichment Factors (EFs) of metals were calculated as EF = (Xaerosol/Refaerosol)/(XUCC/RefUCC) where X is the element under consideration both in aerosol and the upper continental crust (UCC), and Ref is a reference element that is typically crustal such as Al, Fe, Li, Ti etc. Metal abundances in UCC given in Wedepohl (1995) were used and Fe was chosen as the reference metal. Three distinct groups existed among the metals on the basis of their EFs. Manganese was non–enriched in aerosol with mean EFs of 2.6 ± 1.1, 2 ± 0.9 and 1.4 ± 0.7 at RG, MV and MP, respectively. Manganese is typically dominated by crustal contributions, the form found in soil largely being MnO2 (von Schneidemesser et al., 2010). A second group of elements comprising of Cr (16.5 ± 14.6 – 20.6 ± 13.9), Cu (46.4 ± 28.4 – 68.5 ± 38), Ni (56 ± 36.9 – 71.6 ± 50.2) and Pb (67.8 ± 49.3 – 83.9 ± 39.6) had EFs between 10–100 and were moderately–enriched. This is in agreement with published literature as all the elements belonging to this group have a known crustal source in addition to various emission sources (Salvador et al., 2004; Khan et al., 2010). The third group represented by Zn (244.9 ± 155 – 418.3 ± 243) and Cd (386.1 ± 409.8 – 595.3 ± 397.5) had extremely high EF values (between 100–1 000) and were anomalously enriched in aerosol. Vehicular tire wear, battery–manufacture, pigments, metal plating and smelting industries are important sources of these metals (Birmili et al., 2006; Querol et al., 2007). 3.4. Seasonal variation of PM10, metals and influence of meteorological parameters Table 1 shows the seasonal variations of PM10 and associated metals at the study sites. Meteorological parameters recorded in Delhi during the sampling period and the results of correlation analysis between species concentrations and meteorological parameters are provided in Supporting Material (SM) (Tables S1 and S2, respectively). Considering an average aerosol residence time of 5 days, we divided the meteorological data into sets of 5 days (prior to and including the sampling date) to take into account the atmospheric conditions that may have helped in the build–up (or loss) of metal concentrations over a period of time culminating on the sampling date. Overall, seasonal variations of PM10 and ambient metals were highly significant (ANOVA, p<0.001, Table 1), which shows that seasonal conditions possibly influence both the source strengths and the atmospheric accumulation of these species. Levels of PM10
and metals such as Fe and Mn showed a clear seasonal trend of summer>winter>monsoon with significantly higher concentrations in summer as compared to the other seasons (ANOVA with multiple comparisons, p<0.01). The summer season is generally characterized by a high degree of crustal resuspension (of coarser particles) that leads to higher levels of PM10 and crustal metals. Dust storms (“Andhi”) are common in Delhi during summer that carry large amounts of suspended particulate matter (as high as 500–800 tonnes mile–2) from the Thar Desert into Delhi (Singh et al., 2008). Mean temperature, wind speed and the number of bright sunshine hours were positively correlated with the crustal metal Fe (p<0.01, see the SM, Table S2), which is because high solar radiation during summer induces atmospheric convection favoring crustal resuspension and strong winds favor atmospheric dust transport (Querol et al., 2002). Metals such as Cd, Ni, Zn and Cr, on the other hand, had significantly higher concentrations in winter (multiple comparisons, p<0.05) and were negatively correlated with temperature, wind speed and the number of bright sunshine hours. These metals have important anthropogenic sources such as metal smelting (Zn), battery manufacture (Cd, Ni), oil combustion (Ni) (Pacyna and Pacyna, 2001) and vehicular emissions (tailpipe/abrasion) (Zn, Cd, Cr) (Lin et al., 2005; Birmili et al., 2006). Emissions from these sources are trapped and stabilized under inversion layers in winter, leading to higher abundances in ambient aerosol. Relative humidity had significant negative correlations with crustal metals (r = –0.56, –0.53 with Fe and Mn, respectively, p<0.001), which shows that atmospheric moisture enhances settling of suspended particles and prevents crustal resuspension by wetting of soil particles. Understandably, rainfall was negatively correlated with almost all metals as precipitation scavenging is an important mechanism of particulate–removal from the atmosphere (Seinfeld and Pandis, 2006). 3.5. Source apportionment of ambient metals Principal component analysis coupled with multiple linear regression (PCA–MLR) (SPSS 14.0) was used to identify and quantify sources of atmospheric metals. The multiple linear model used here is the one described in Larsen and Baker (2003)
(1)
where PCA factor scores serve as the independent variables (Xi) and the total metal concentration in aerosol serves as the dependent variable (y). The influence of each independent variable on the dependent variable can be directly compared by the regression coefficients, if the dependent and independent variables are normally standardized. When the variables of Equation (1) are normalized, the regression coefficients are
Khillare and Sarkar – Atmospheric Pollution Research 3 (2012) 46‐54
50
represented as B, and the intercept (b) is zero. The equation becomes: (2)
where z is the standardized normal deviate of the total metal concentration in aerosol (the dependent variable), Bi is the partial regression coefficient and Xi is the PCA factor score for source i (the independent variable). By expanding z and rearranging terms, we get the following multiple linear regression equation: (3)
where TM is the total metal concentration in aerosol, TM* is the mean total metal concentration, σTM is the corresponding standard deviation and FSi is the factor score for source i. The mean contribution of source i can then be calculated as:
%
100
Σ
(4)
and daily contributions can be calculated as:
Σ
(5)
where Bi/ΣBi is the ratio of the regression coefficient for factor i to the sum of all the regression coefficients. PCA factors were identified using Varimax rotation with Kaiser Normalization. Eigenvalue>1 was the criterion for selecting factors and a factor score of 0.5 was selected as the lowest level of signif– icance within a factor. PCA factor scores for individual sites are presented in Table 3. Three factors explained 78.4% variance of the data at RG. The first factor (PC1) was loaded with Cd, Cu, Ni and Pb representing impacts of industrial sources present in the upwind direction (possibly in the form of metal smelting, metal product manufacture and industrial oil combustion) (Querol et al., 2007; Kulshrestha et al., 2009). PC2 was dominated by Fe, Mn and Cr suggesting crustal origin (Shridhar et al., 2010). PC3 was dominated by Zn and moderately loaded with Mn. Zinc is a known tracer of tire wear particles (Birmili et al., 2006; Wang et al., 2006). Manganese was predominantly crustal in this study (see Section 3.3) indicating that its loading in this factor was probably due to resuspension of road dust by vehicular turbulence. So, this factor represents vehicular sources.
Three factors were identified that explained 75.5% data variance at MV. PC1 was loaded with Cd, Ni, Pb, Zn and Cr, possibly indicating industrial emission sources (Querol et al., 2007; Kulshrestha et al., 2009) from the Patparganj Industrial Estate. PC2 was highly loaded with Mn, Cd and Cu, and to a lesser extent with Fe, denoting vehicular resuspension of road dust and brake–drum abrasion particles (Birmili et al., 2006; Wang et al., 2006). PC3 was loaded with Fe and Pb indicating resuspension of crustal and road– dust possibly containing residual Pb–additives. Three factors explained 74.1% variance of the dataset at MP. PC1 was significantly loaded Cd, Ni, Zn and Cr, suggesting industrial emissions possibly originating from the Okhla Industrial Area present upwind of the site. PC2 was dominated by Cu, Pb and Cr suggesting emissions from brake–drum abrasion (Wang et al., 2006) and resuspension of road dust by vehicular turbulence. PC3 had high loadings of Fe and Mn indicating a crustal source. Observed and PCA–MLR–predicted total metal concentrations agreed strongly (0.80
Table 3. Results of Principal Component Analysis (varimax rotation with Kaiser Normalization) at the individual sites. Only factor loadings ≥0.3 are shown. Loadings ≥0.5 are in bold
Species Fe Mn Cd Cu Ni Pb Zn Cr Eigenvalues % of variance Cumulative %
RG Principal Components PC1 PC2 PC3 0.92 0.67 0.43 0.88 0.85 0.81 0.35 0.86 0.94 0.39 0.67 3.13 2 1.13 39.14 25.11 14.11 78.35 39.14 64.25
MV Principal Components PC1 PC2 PC3 0.36 0.79 0.82 0.54 0.60 0.77 0.80 0.52 0.70 0.82 0.33 0.91 2.71 1.83 1.51 33.82 22.83 18.89 33.82 56.65 75.54
MP Principal Components PC1 PC2 PC3 0.30 0.86 0.84 0.82 0.43 0.65 0.71 0.88 0.36 0.80 0.30 0.57 0.64 2.38 1.9 1.65 29.7 23.74 20.65 29.7 53.45 74.1
Khillare and Sarkar – Atmospheric Pollution Research 3 (2012) 46‐54
51
–3
Figure 2. Time–evolution of source contributions (µg m ) at the study sites (a); mean annual contribution of apportioned sources at the study sites (b); agreements between observed and model–predicted total metal concentrations at the sites (c).
(Planning Department, 2009). With this trend continuing, the vehicular source could become even more important in the coming years. Industrial emissions accounted for 4–21% of the analyzed metal mass and showed significant spatial (F = 116.8, p<0.001) and seasonal variation (F = 39.1, p<0.001). The moderate distance of these sites from the nearest industrial areas (3–10 km) required that the emitted pollutants be atmospherically transported from the source to the receptor, thus depending heavily on seasonal variations in atmospheric conditions. Most of the small–scale industries located in Delhi cannot afford pollution abatement technologies and pose severe problems for effective monitoring and enforcement. 3.6. Potential health risks to the exposed population Theoretical estimates of blood lead levels in children. Ambient Pb concentrations (0.27 ± 0.2, 0.41 ± 0.2 and 0.46 ± 0.3 µg m–3 at RG, MV and MP, respectively) were lower the NAAQS (0.5 µg m–3, annual average). Airborne Pb concentrations in India have declined significantly since the complete phase–out of leaded gasoline in 1999; however, the levels found in this study are still higher than many other cities (Table 2). The relationship between blood lead (PbB) of children and air lead (PbA) can only be weakly estimated because of a number of factors (Brunekreef, 1984): firstly, since children spend only a small fraction of their time outdoors, Pb concentration measured in outdoor air is poorly correlated to personal Pb exposure; secondly, PbB shows large variability near lead emitting sources such as smelters whereas in urban areas the variability is much less; lastly, environmental factors varying between studies also influences the
relationship to a large extent. Despite this variability, Brunekreef (1984) compiled data from 96 different child populations of varying ages, from different countries, from 18 different study areas and found a statistically significant relationship between PbB and PbA as ln PbB = 2.8528 + 0.3485 ln PbA (R2 = 0.692 after log–transfor– mation). This relationship may tend to overestimate PbB at high levels of exposure and upon restricting the analysis to populations with PbB<20 µg dL–1 (N = 43), a second relationship was suggested as a more conservative estimate in the form of ln PbB = 2.607 + 0.2159 ln PbA (R2 = 0.331) (Brunekreef, 1984; Browne et al., 1999). We used these two equations as upper and lower estimates, respectively, and found estimated PbB in children living near RG, MV and MP sites to be 10.2–10.9, 11.2–12.7 and 11.5–13.3 µg dL–1, respectively. We also compared these estimates with actual PbB data of children reported by various authors from India (specifically, from Delhi, Mumbai and Chennai) to check the applicability of the equations in the Indian scenario. Jain and Hu (2006) studied PbB levels in 1 081 children belonging to Mumbai and Delhi and found that most of the children (76%) had PbB values between 5 and 20 µg dL–1. Kalra et al., (2003) studied PbB levels of 190 children (125 school children and 65 children belonging to an urban slum) in Delhi and found a mean value of 7.8 ± 3.9 µg dL–1 while Roy et al., (2009) found a mean PbB level of 11.4 ± 5.3 µg dL–1 among 756 children in Chennai. It is apparent that our estimates agree very well with the actual PbB data reported by these authors, which means that the representa– tiveness of the equations in the present case is well–justified. The United States Center for Disease Control (CDC) defines elevated blood lead levels as ≥10 µg dL–1 (CDC, 1991) and recommends that children with blood lead ≥20 µg dL–1 should be referred for medical evaluation. However, adverse health effects have been observed
Khillare and Sarkar – Atmospheric Pollution Research 3 (2012) 46‐54
52
for concentrations below 10 µg dL–1 (Canfield et al., 2003). For example, Roy et al., (2009) found that a mean PbB level of 11.4 ± 5.3 µg dL–1 in children of Chennai was associated with higher anxiety, social problems and greater attention–deficit hyperactivity disorder (ADHD). Our estimates are quite comparable to these values and it seems that children living in the study areas might suffer from some degree of lead toxicity. Incremental lifetime cancer risk assessment. The overall annual averages of Ni and Cd in this study (316 ± 161 and 14 ± 12 ng m–3, respectively) were around 15 and 3 times higher than the stipulated standards (NAAQS for Ni: 20 ng m–3 and the European Union target for Cd: 5 ng m–3, EU Directive 2004/107/CE). As per US EPA’s weight–of–evidence, Ni is classified as a Group A pollutant (known human carcinogen) while Cd is classified as a Group B1 pollutant (probable human carcinogen) (IRIS, 1995). Hexavalent Cr – unregulated in outdoor air – is also carcinogenic to humans and is classified as a Group A pollutant (IRIS, 1995). Incremental lifetime cancer risk (ILCR) is the incremental probability of an individual developing cancer over a lifetime (70 years) as a result of exposure to a potential carcinogen. ILCR was calculated based on the method described by Gurjar and Mohan (2003): (6) –1 where, CDI is the chronic daily intake (mg kg day) and ISF is the inhalation slope factor (mg kg–1 day)–1. CDI can be further calculated as:
(7)
where, CC is the contaminant concentration (mg m–3), IR is the inhalation rate (m3 day–1), EF is the exposure frequency (days yr–1), ED is the exposure duration (yr), AF is the absorption fraction, BW is the body weight (kg), and LT is the average lifetime (days) i.e. age (yr) × 365 days yr–1. The parameters used in the calculation of ILCR and results of the assessment are presented in Table 4. Concentration of Cr (VI) was assumed to be no less than one–seventh of total Cr (Mancuso, 1975). This serves as a conservative estimate of Cr (VI) and has been used by various recent studies involving risk characterization (Pratt et al., 2000; Gurjar and Mohan, 2003; DEP, 2009). Calculated ILCRs for Cd, Cr (VI) and Ni were 6 × 10–6, 1.5 × 10–4 and 1.5 × 10–5, respectively. An ILCR between 10–6 and 10–4 is generally considered –4 to indicate potential risk while ILCR greater than 10 suggests high potential health risk. In comparison, the risks estimated in this study are somewhat higher for Cr (VI) and Ni. Societal ILCR was calculated assuming Delhi’s population to be 17 million and it was found that up to 2 908 excess cancer cases (102 for Cd, 2 559 for Cr (VI) and 247 for Ni) are possible in Delhi considering lifetime inhalation exposure to these pollutants at their current concen– trations.
4. Conclusion Eight major and trace metals associated with PM10 were characterized at three residential areas in Delhi, India to identify and quantify their major sources. PCA–MLR identified dominance of a crustal source (49–65%) having significantly higher contributions in summer. Local sources such as construction activities, road dust resuspension etc. coupled with trans– boundary migration of dust most likely contributed to this factor. Given the impossibility to control this source, the only viable option left is to prevent resuspension of dust to any feasible degree. Various developed countries have successfully used
Table 4. Incremental lifetime cancer risk (ILCR) assessment for exposure to airborne Cd, Cr (VI) and Ni at their respective mean annual concentrations observed in this study Species Concentration –3
(mg m ) Cd
c
–1
(mg kg day)
CDI –1
ILCR
–1
Societal e ILCR
–6
(mg kg day) (10 )
–5
6.1
9.86*10
–5
41
3.67*10
–4
0.84
1.73*10
15
247
Total
198
2 908
1.44*10 a
b
ISF
Cr (VI)
2.68*10
Ni
3.16*10
–7
6
102
–6
151
2 559
–5
a
Concentration of Cr (VI) has been calculated as 1/7 of total chromium (Mancuso, 1975). b Inhalation slope factor (Source: The Risk Assessment Information System; http://rais.ornl.gov) c Chronic daily intake; parameters used for calculating CDI are: Species –3 3 –1 concentration (mg m ), Inhalation rate (IR) = 20 m day , Exposure –1 frequency (EF) = 350 days yr , Exposure duration (ED) = 70 yrs, Absorption d d d fraction = 0.25 for Cd , 0.5 for Cr (VI) and 0.2 for Ni , Body weight (BW) = 70 kg, Average lifetime (LT) = 70 yrs × 365 days i.e. 25 550 days. d Gurjar and Mohan (2003). e –6 6 Societal ILCR = ILCR (10 ) × population (10 ).
measures such as large–scale water flushing and chemical suppression to prevent resuspension of road dust (Amato et al., 2010). However, the practical aspects of using such methods in India are yet to be looked into. Vehicular sources contributed 27– 31% to ambient metals and included tailpipe/abrasion emissions along with resuspended dust. This source is likely to be even more significant in the finer fraction of ambient particles (PM2.5 and less). Industrial emissions (metal smelting and manufacture, oil combustion etc.) contributed 4–21% at the sites. Overall, it was found that residential areas of Delhi are significantly more polluted than their counterparts in other countries. Possible health risks include high estimated blood lead levels (10.2–13.3 µg dL–1) in children and around 2 908 excess cancer cases (102 for Cd, 2 559 for Cr (VI) and 247 for Ni) that are likely in Delhi considering lifetime inhalation exposure to these pollutants at their observed concentrations.
Acknowledgements Sayantan Sarkar wishes to thank Indian Association of Parliamentarians on Population and Development (IAPPD) for providing financial assistance in the form of Sat Paul Mittal Fellowship during the course of this work. This work was financially supported by a project (No. 19/4/2007–RE) sponsored by the Ministry of Environment and Forests (MoEF), Government of India. The authors gratefully acknowledge the anonymous reviewers for their critical comments and constructive suggestions.
Supporting Material Available Meteorological parameters recorded in Delhi during the study period (Table S1), Correlation coefficients observed between analyzed metal concentrations and 5–day averaged meteorological parameters (Table S2). This information is available free of charge via the Internet at http://www.atmospolres.com.
References Amato, F., Querol, X., Johansson, C., Nagl, C., Alastuey, A., 2010. A review on the effectiveness of street sweeping, washing and dust suppressants as urban PM control methods. Science of the Total Environment 408, 3070‐3084. Birmili, W., Allen, A.G., Bary, F., Harrison, R.M., 2006. Trace metal concentrations and water solubility in size‐fractionated atmospheric particles and influence of road traffic. Environmental Science and Technology 40, 1144‐1153.
Khillare and Sarkar – Atmospheric Pollution Research 3 (2012) 46‐54
53
Browne, D.R., Husni, A., Risk, M.J., 1999. Airborne lead and particulate levels in Semarang, Indonesia and potential health impacts. Science of the Total Environment 227, 145‐154.
Limbeck, A., Handler, M., Puls, C., Zbiral, J., Bauer, H., Puxbaum, H., 2009. Impact of mineral components and selected trace metals on ambient PM10 concentrations. Atmospheric Environment 43, 530‐538.
Brunekreef, B., 1984. The relationship between air lead and blood lead in children: a critical‐review. Science of the Total Environment 38, 79‐123.
Lin, C.C., Chen, S.J., Huang, K.L., Hwang, W.I., Chang‐Chien, G.P., Lin, W.Y., 2005. Characteristics of metals in nano/ultrafine/fine/coarse particles collected beside a heavily trafficked road. Environmental Science and Technology 39, 8113‐8122.
Cadle, S.H., Mulawa, P.A., Hunsanger, E.C., Nelson, K., Ragazzi, R.A., Barrett, R., Gallagher, G.L., Lawson, D.R., Knapp, K.T., Snow, R., 1999. Composition of light‐duty motor vehicle exhaust particulate matter in the Denver, Colorado area. Environmental Science and Technology 33, 2328‐2339. Canfield, R.L., Henderson, C.R., Cory‐Slechta, D.A., Cox, C., Jusko, T.A., Lanphear, B.P., 2003. Intellectual impairment in children with blood lead concentrations below 10 µg per deciliter. New England Journal of Medicine 348, 1517‐1526. CDC (Centers for Disease Control), 1991. Preventing lead poisoning in young children: a statement by the Centers for Disease Control. Public Health Service, US Department of Health and Human Services, Atlanta, GA. pp.1‐137. Contini, D., Genga, A., Cesari, D., Siciliano, M., Donateo, A., Bove, M.C., Guascito, M.R., 2010. Characterisation and source apportionment of PM10 in an urban background site in Lecce. Atmospheric Research 95, 40‐54. de Kok, T.M.C.M., Driece, H.A.L., Hogervorst, J.G.F., Briede, J.J., 2006. Toxicological assessment of ambient and traffic‐related particulate matter: a review of recent studies. Mutation Research‐Reviews in Mutation Research 613, 103‐122. DEP (Pennsylvania Department of Environmental Protection), 2009. DEP sampling study at USA TODAY report sites. Pennsylvania Department of Environmental Protection, Reading, PA. Forsberg, B., Hansson, H.C., Johansson, C., Areskoug, H., Persson, K., Jarvholm, B., 2005. Comparative health impact assessment of local and regional particulate air pollutants in Scandinavia. Ambio 34, 11‐19.
Mancuso, T.F., 1975. Consideration of chromium as an industrial carcinogen. International conference on heavy metals in the environment, October 27‐ 31, 1975, Toronto, Ontario, Canada, 343– 356. Meresova, J., Florek, M., Holy, K., Jeskovsky, M., Sykora, I., Frontasyeva, M.V., Pavlov, S.S., Bujdos, M., 2008. Evaluation of elemental content in air‐borne particulate matter in low‐level atmosphere of Bratislava. Atmospheric Environment 42, 8079‐8085. MoEF (Ministry of Environment and Forests), 2009. Environment (Protection) Seventh Amendment Rules,Government of India Press, New Delhi. Nawrot, T.S., Kuenzli, N., Sunyer, J., Shi, T.M., Moreno, T., Viana, M., Heinrich, J., Forsberg, B., Kelly, F.J., Sughis, M., Nemery, B., Borm, P., 2009. Oxidative properties of ambient PM2.5 and elemental composition: heterogeneous associations in 19 European cities. Atmospheric Environment 43, 4595‐4602. Pacyna, J.M., Pacyna, E.G., 2001. An assessment of global and regional emissions of trace metals to the atmosphere from anthropogenic sources worldwide. Environmental Reviews 9, 269–298. Paulino, S.A., Quiterio, S.L., Escaleira, V., Arbilla, G., 2010. Evolution of particulate matter and associated metal levels in the urban area of Rio de Janeiro, Brazil. Bulletin of Environmental Contamination and Toxicology 84, 315‐318. Planning Department, 2009. Economic survey of Delhi 2008–2009. Government of NCT of Delhi.
Gurjar, B.R., Mohan, M., 2003. Potential health risks due to toxic contamination in the ambient environment of certain Indian states. Environmental Monitoring and Assessment 82, 203‐223.
Pope, C.A., Dockery, D.W., 2006. Health effects of fine particulate air pollution: lines that connect. Journal of the Air and Waste Management Association 56, 709‐742.
Gurjar, B.R., Jain, A., Sharma, A., Agarwal, A., Gupta, P., Nagpure, A.S., Lelieveld, J., 2010. Human health risks in megacities due to air pollution. Atmospheric Environment 44, 4606‐4613.
Pratt, G.C., Palmer, K., Wu, C.Y., Oliaei, F., Hollerbach, C., Fenske, M.J., 2000. An assessment of air toxics in Minnesota. Environmental Health Perspectives 108, 815‐825.
Heal, M.R., Hibbs, L.R., Agius, R.M., Beverland, L.J., 2005. Total and water‐ soluble trace metal content of urban background PM10, PM2.5 and black smoke in Edinburgh, UK. Atmospheric Environment 39, 1417‐1430.
Querol, X., Viana, M., Alastuey, A., Amato, F., Moreno, T., Castillo, S., Pey, J., de la Rosa, J., de la Campa, A.S., Artinano, B., Salvador, P., Dos Santos, S.G., Fernandez‐Patier, R., Moreno‐Grau, S., Negral, L., Minguillon, M.C., Monfort, E., Gil, J.I., Inza, A., Ortega, L.A., Santamaria, J.M., Zabalza, J., 2007. Source origin of trace elements in PM from regional background, urban and industrial sites of Spain. Atmospheric Environment 41, 7219‐7231.
IARC (International Agency for Research on Cancer), 2009. Complete list of agents evaluated and their classification. International Agency for Research on Cancer. http://monographs.iarc.fr/ENG/Classification/ index.php. IRIS (Integrated Risk Assessment System), 1995. United States Environmental Protection Agency, www.epa.gov/IRIS/. Jain, N.B., Hu, H., 2006. Childhood correlates of blood lead levels in Mumbai and Delhi. Environmental Health Perspectives 114, 466‐470.
Querol, X., Alastuey, A., de la Rosa, J., Sanchez‐de‐la‐Campa, A., Plana, F., Ruiz, C.R., 2002. Source apportionment analysis of atmospheric particulates in an industrialised urban site in southwestern Spain. Atmospheric Environment 36, 3113‐3125.
Jorquera, H., 2009. Source apportionment of PM10 and PM2.5 at Tocopilla, Chile (22°05'S, 70°12'W). Environmental Monitoring and Assessment 153, 235‐251.
Roy, A., Bellinger, D., Hu, H., Schwartz, J., Ettinger, A.S., Wright, R.O., Bouchard, M., Palaniappan, K., Balakrishnan, K., 2009. Lead exposure and behavior among young children in Chennai, India. Environmental Health Perspectives 117, 1607‐1611.
Kalra, V., Chitralekha, K.T., Dua, T.R., Pandey, R.M., Gupta, Y., 2003. Blood lead levels and risk factors for lead toxicity in children from schools and an urban slum in Delhi. Journal of Tropical Pediatrics 49, 121‐123.
Salvador, P., Artinano, B., Alonso, D.G., Querol, X., Alastuey, A., 2004. Identification and characterisation of sources of PM10 in Madrid (Spain) by statistical methods. Atmospheric Environment 38, 435‐447.
Khan, M.F., Hirano, K., Masunaga, S., 2010. Quantifying the sources of hazardous elements of suspended particulate matter aerosol collected in Yokohama, Japan. Atmospheric Environment 44, 2646‐2657.
Sanchez de la Campa, A.M., de la Rosa, J., Querol, X., Alastuey, A., Mantilla, E., 2007. Geochemistry and origin of PM10 in the Huelva region, Southwestern Spain. Environmental Research 103, 305–316.
Kulshrestha, A., Satsangi, P.G., Masih, J., Taneja, A., 2009. Metal concentration of PM2.5 and PM10 particles and seasonal variations in urban and rural environment of Agra, India. Science of the Total Environment 407, 6196‐6204. Larsen, R.K., Baker, J.E., 2003. Source apportionment of polycyclic aromatic hydrocarbons in the urban atmosphere: a comparison of three methods. Environmental Science and Technology 37, 1873‐1881.
Schwartz, J., Dockery, D.W., Neas, L.M., 1996. Is daily mortality associated specifically with the particles? Journal of the Air and Waste Management Association 46, 927–939. Seinfeld, J.H., Pandis, S.N., 2006. Atmospheric Chemistry and Physics: From nd Air Pollution to Climate Change. 2 ed., John Wiley and Sons, New York.
54
Khillare and Sarkar – Atmospheric Pollution Research 3 (2012) 46‐54
Shridhar, V., Khillare, P.S., Agarwal, T., Ray, S., 2010. Metallic species in ambient particulate matter at rural and urban location of Delhi. Journal of Hazardous Materials 175, 600‐607. Singh, M., Jaques, P.A., Sioutas, C., 2002. Size distribution and diurnal characteristics of particle‐bound metals in source and receptor sites of the Los Angeles basin. Atmospheric Environment 36, 1675‐1689. Singh, T., Khillare, P.S., Shridhar, V., Agarwal, T., 2008. Visibility impairing aerosols in the urban atmosphere of Delhi. Environmental Monitoring and Assessment 141, 67‐77. Srivastava, A., Jain, V.K., 2008. Source apportionment of suspended particulate matters in a clean area of Delhi: a note. Transportation Research Part D‐Transport and Environment 13, 59‐63. Stone, E., Schauer, J., Quraishi, T.A., Mahmood, A., 2010. Chemical characterization and source apportionment of fine and coarse particulate matter in Lahore, Pakistan. Atmospheric Environment 44, 1062‐1070. Theodosi, C., Im, U., Bougiatioti, A., Zarmpas, P., Yenigun, O., Mihalopoulos, N., 2010. Aerosol chemical composition over Istanbul. Science of the Total Environment 408, 2482‐2491.
Vijayanand, C., Rajaguru, R., Kalaiselvi, K., Selvam, K.P., Palanivel, M., 2008. Assessment of heavy metal contents in the ambient air of the Coimbatore city, Tamilnadu, India. Journal of Hazardous Materials 160, 548‐553. von Schneidemesser, E., Stone, E.A., Quraishi, T.A., Shafer, M.M., Schauer, J.J., 2010. Toxic metals in the atmosphere in Lahore, Pakistan. Science of the Total Environment 408, 1640‐1648. Wang, X.L., Sato, T., Xing, B.S., 2006. Size distribution and anthropogenic sources apportionment of airborne trace metals in Kanazawa, Japan. Chemosphere 65, 2440‐2448. Wedepohl, K.H., 1995. The composition of the continental‐crust. Geochimica Et Cosmochimica Acta 59, 1217‐1232. WHO (World Health Organization), 2006. WHO air quality guidelines for particulate matter, ozone, nitrogen dioxide and sulfur dioxide: Global update 2005. Yatkin, S., Bayram, A., 2010. TSP, PM depositions, and trace elements in the vicinity of a cement plant and their source apportionments using chemical mass balance model in Izmir, Turkey. Environmental Monitoring and Assessment 167, 125‐141.